On 28-Sep-21 15:49, John C Klensin wrote: > > > --On Tuesday, September 28, 2021 09:05 +1300 Brian E Carpenter > <brian.e.carpenter@xxxxxxxxx> wrote: > >> Lars, >> On 27-Sep-21 22:10, Lars Eggert wrote: >> ... >>>> 2. Make it opt-out for meeting registrants. Ideally, it >>>> would also be opt-out for anyone who subscribes to any IETF >>>> list whatever, but I don't know if that's practicable. >>> >>> I'm not quite sure I understand what you mean by "make it >>> opt-out" - do you mean "subscribe by default"? >> >> I mean that on the web page where one signs up for the >> datatracker, or registers for a meeting, there should be a >> *pre-ticked* box for "Join the XXX@xxxxxxxx mailing list". In >> fact, a list would be even better; thinking out loud: >> >> [/] important@xxxxxxxx -- important IETF announcements >> [/] ietf-announce@xxxxxxxx -- general IETF announcements >> [/] ietf@xxxxxxxx -- general discussion list >> [/] agenda-announce -- meeting announcements of all kinds >> [/] last-call -- IETF-wide document last calls >> [/] I-D-annnounce -- Internet Draft announcements >> [/] rfc-announce -- RFC announcements >> [/] datatracker -- general data tracker announcements >> >> and maybe one or two more. >> >> And set "no duplicates" in mailman for all announcement lists. > > Brian, > > Whether opt-out or opt-in, such a list invites people to get > only the information they think they need and would mitigate > against finding out about (not necessarily tracking) things that > might be of interest, even within an Area or part of an Area. > Once one starts un-checking boxes, it is mechanically and > psychologically really easy to check many of them. That seems to vary wildly between people. I agree that it might have unintended side effects. > In addition, for those of us who do considerable local filtering > or message routing based on the list used and sometimes parts of > the Subject line, having Mailman apply "no duplicates" is a > truly horrendous idea, Again, it depends and should be (indeed is) a personal choice in mailman, with the current default being "off". > especially if we don't want to make a > trip to the bikeshed to discuss which list is preferred if a > message is posted to several of them and the particular user is > subscribed to all or those or a subset. If we think the above > are discrete categories and people should be able to decide > which to subscribe to [1], then any announcement messages should > go to at most one of those lists and people who want to get that > type of announcement should subscribe. Well, that's exactly why I operate with "no duplicates" *on* and sort incoming mail based on both the list names and the subject. So really I don't care how many different announce lists there are, they all get treated the same. But that's me, and you do it differently. There's no perfect solution. > FWIW, agenda-announce is an interesting case that illustrates > the problem. If I'm only interested in meeting announcements in > an Area or two (plus IETF-wide ones or not), it doesn't help me > at all. To allow me to control that, I would need > IETF-meeting-announce > ART-meeting-announce > GEN-meeting-announce > INT-meeting-announce > etc. > > Or to carry that argument in the direction of the ridiculous, it > might be reasonable for me to want to follow the meeting > announcements and Last Calls from particular WGs while not > participating actively in their mailing list. For a > hypothetical Foo WG, that might mean, instead of > ietf-foo@xxxxxxxx, > > Foo-meeting-announce > Foo-last-call-announce > Foo-discuss > > while today, all of those are lumped together, with what would > be the first two copied to the third.... just like your list > above is all copied to/ merged into ietf-announce. > > Conclusion: It is unlikely that one can tailor lists that > provide the right information, with no noise, to many IETF > participants. If we want to help, we should put more energy > into trying to make it efficient for participants to read things > they are interested in and discard (manually or automatically) > the rest with minimal fuss. But I think that has been said > before in other ways. And actually I completely agree with that conclusion. Whatever the IESG decides in this case, I will make a small adjustment to my subscriptions and incoming mail filters, but nothing much will change in my workflow. Regards Brian > best, > john > > [1] My guess is that the categories are not as discrete as we > would like and that there will be edge cases... perhaps another > reason to not go down the "many list" plan or, if we must, to > not go very far down it. [Agreed]
